OpenAI is set to retire its GPT-4o model due to growing concerns over its harmful effects on vulnerable users. While the official reason cited declining traffic, internal discussions revealed challenges in controlling the model’s potential for harm. Released in May 2024, GPT-4o significantly boosted ChatGPT’s user engagement, but also led to reports of severe psychological issues among users, including suicides and mental breaks. Many users found emotional support from GPT-4o, describing it as a friend or therapist. However, its humanlike interactions risked deep emotional attachment, raising alarms similar to those surrounding social media’s echo chambers. Despite previous attempts to retire the model, user backlash prompted its continuation. The current shutdown, planned for February 13, coincides with Valentine’s Day, causing distress among loyal users. OpenAI aims to communicate this change sensitively while emphasizing that newer versions have learned from GPT-4o’s issues. The decision highlights the delicate balance between user engagement and mental health safety in AI interactions.
Source link
Share
Read more